This project is developing a system that allows manipulation of complex displays in real time, contingent on the eye position of an observer with a free head. This is crucial for understanding how the successive views of a scene generated by observer motion are related. Achieving an appropriate relationship between successive retinal images is fundamental for understanding the reference frames underlying (a) perception of a stable visual world, (b) short- and long-term visual spatial memory, and (c) much of our motor interaction with the world. There are three major classes of questions we can investigate making saccade-contingent display changes: (1) What is the nature of the visual information retained from previous views? (2) What are the reference frames for programming the various eye/head movements in the task? (3) What is the nature of the subcomponents of the task?